/learning-reactive-islands

Machine learning for reaction dynamics. Package contains support vector machines for learning reactive islands, computing tube manifolds for Hénon-Heiles system

Primary LanguageJupyter Notebook

Machine learning for phase space structures in reaction dynamics

Collection of MATLAB, Python scripts, and notebooks to generate training data and support vector classifier models for learning reactive islands in a model Hamiltonian. The model Hamiltonian is the two degrees of freedom H{'e}non-Heiles system which is an open non-linear dynamical system and considered as a benchmark for method development and verification.